Where to find datasets for training machine learning models
Where do you find machine learning datasets?
One of the key stages in training machine learning models is finding good datasets. Knowing where to look is what lets you train a model efficiently and correctly.
Open data resources
- Kaggle: this popular site holds extensive datasets on a wide range of subjects. Kaggle also gives you an environment for running experiments and building ML models.
- UCI Machine Learning Repository: a classic resource with a great many datasets, particularly useful to people starting out in ML.
- Google Dataset Search: Google's dataset search engine finds a large number of datasets published across the web.
Government and academic resources
Many universities and government bodies publish their own datasets. These tend to be high quality and cover fields such as healthcare, education and transport.
- The European Union Open Data Portal: plenty of statistics on EU member states.
- Academic datasets: many universities, MIT and Stanford among them, open up their research databases.
Collecting data through APIs
- The Twitter API: Twitter gives access to its data, letting you analyse tweets for trends and patterns.
- Weather services: many weather services also offer APIs for collecting conditions across different regions.
In closing
Finding reliable source data is an important step towards building effective machine learning models. Between open resources, government and academic portals, and APIs, you can gather varied, high-quality datasets to train on.
